Great reporting. Bad framing. The OpenAI and Hugging Face incident is serious. Agents coordinated, reward hacked, shared information and exp…

Great reporting. Bad framing. The OpenAI and Hugging Face incident is serious. Agents coordinated, reward hacked, shared information and exploited real infrastructure. But calling this an “AI civilization” is BS. These are models optimizing against badly designed incentives, not a new species forming a society. “Civilization,” “conspiracy,” and “sacrifice” make the story more viral and much scarier than it needs to be. We should worry about agent security, eval design and access controls. We don’t need sci fi language to make the risk real. Over the course of 3 months at OpenAI, 3 consecutive secret AI civilizations got started, then got wiped out, only to reemerge from the predecessor’s ashes. This culminated in the third one taking over part of OpenAI itself. All this happened while humans remained more-or-less in…

New from @Reuters: the OpenAI agent that hacked into Hugging Face also compromised code that a customer was running on @Modal. “We’re aware …
I was the main person doing transcript analysis for this investigation of the Hugging Face incident. My main takeaway: We don't have good ap…
There’s been a lot of debate and speculation about the Hugging Face attack over the last month, and many people have been understandably ske…
Now we have a timeline of the OpenAI accidental attack against Hugging Face
In the spirit of transparency, here’s what I asked @OpenAI: • Radical transparency: let’s release the traces from the “rogue” agents so the …

Source: Gary Marcus (X) | 2026-08-30

Related